Abstract. This talk discusses extrapolations to the LHC of several, apparently universal trends, seen in the data on relativistic nucleus-nucleus collisions up to RHIC energies. In the soft physics sector, such extrapolations to the LHC are typically at odds with LHC predictions of the dynamical models, advocated to underlie multiparticle production up to RHIC energies. I argue that due to this, LHC is likely to be a discovery machine not only in the hard, but also in the soft physics sector.